Mental Models

• The Adjacent Possible discoveries happen at the frontier, find what has just become possibe or is just barely possible and look for natural next steps
• train yourself to see subtle differences between apparently similar concepts: often how important breakthroughs happen
• Bias Variance Trade off
• Lepton Fake Factor / ABCDMethod : estimate the unknown by extrapolating from data control regions using simple assumption
• Unit Tests or Closure Tests: verify a method gives the right answer on a known case before trusting it on the unknown
• Circle of Competencies know what you don't know; probe where an expert's authority actually ends before trusting it
• First principles thinking strip away assumptions and analogies; rebuild reasoning from foundational truths
• The map is not the territory models are useful but necessarily incomplete; don't confuse the abstraction for reality
• Law of diminishing returns each additional unit of input yields less output than the last
• 2nd order thinking ask what happens next after the first-order effect; most people stop one step too early
• Inversion: Try to avoid failure rather than achieve success
• Occams Razor among competing explanations, prefer the one with fewest assumptions
• Hanlons Razor never attribute to malice what can be explained by negligence
• Abstractions can be useful focus on the relevant degrees of freedom
• See The Good Sturgeon's Law antidote — 90% is crap, train yourself to find the 10% worth keeping
• Free roll: decisions with asymmetry between upside and downside b/c potential losses are insignificant
• Applause light Saying something that is so trite that the reversal sounds abnormal, then the unreversed statement is not conveying new information.
• Slow is smooth and smooth is fast rushing causes errors that cost more time than slowing down would have
• Feelings are short cuts to effective thinking emotions are evolved probability estimators, not noise; compressed evolutionary rationality
• Compounding returns small consistent gains multiply; the return on the return eventually dominates
• Absence Blindness: Humans have a hard time paying attention to what is NOT present
• Features of complex systems are often a result of their developmental path not a design
• Build structures around autonomy design systems that protect individual freedom rather than constrain it; libertarian org principle
• It pays to get the design right clever structure tuned to the problem dramatically outperforms brute force
• Long chains of complex reasoning or complicated plans are brittle each link multiplies failure probability; prefer short robust chains
• Take the other persons point of view genuinely model their perspective before responding
• Incentives Matter
• Hard decisions are easy when a decision feels hard, the options are nearly equal — closeness is a signal to decide fast, not slow down
• Decisions Stacking: Make lower-impact, easier-to-quit decisions that teach you something before the high-impact one-way decision
• The Middle Way
• Generator Filter adversarial pair creates dynamic stability — one generates, one filters; GANs, separation of powers, minimax
• Context matters the same fact means different things in different contexts; discovery is easy with the right context
• The Bentham Horizon: blind spot of pure reductionism; emergent/emotional reality lives beyond what can be broken into parts
• Sunk Cost Fallacy: past investment is irrelevant to future decisions; only future costs and benefits matter
• Survivorship Bias: we study winners because losers are invisible; conclusions from survivors are systematically overoptimistic
• Pre-mortem: before committing, imagine the plan has failed and work backwards to find the most likely cause
• Reference Class Forecasting (How Big Things Get Done)
• Goodhart's Law — when a measure becomes a target it ceases to be a good measure.
• Chesterton's Fence — don't remove something until you understand why it's there.
• Local vs Global Optima: gradient descent finds nearby peaks, not the best peak; escaping local optima requires stepping back
• Opportunity Cost: choosing one thing means forgoing the next-best alternative; the real cost is what you didn't do
• Skin in the Game: decision-makers who bear consequences make better decisions; absence of exposure corrupts judgment
• Lindy Effect: the longer something has survived, the longer it's likely to keep surviving
• Antifragility: some systems gain from disorder rather than merely surviving it; volatility is fuel not just risk
• Jevons Paradox: making energy use more efficient increases total energy consumption.
• Thought Terminating Jargon - Abstractions and Formalisms allow us to stop thinking and give into our preconceived biases. We are confident that the other side is why, without understanding why.